Seminar on Closed Loop Medical Devices (CSCI 7000-015, Spring 2016)
Modeling, Design and Verification of Medical Cyber-Physical Systems
Course Information
Class Timings: Wednesday 11AM - 12 Noon.
Class Location: ECCR 139
Instructor: Sriram Sankaranarayanan
Pre-Requisites: Consent of the instructor.
Office Hours: TBA
Final Exam: No Exam.
Course Format
Each week, we will read and discuss a group of related papers on the
topic of closed loop medical devices. For some of the weeks a student
will be designated to lead the discussion. The goal for each student
is to read the paper as carefully as possible before the week's
lecture and contribute to its discussion in class.
Before Class
A discussion lead for that week is assigned in advance. Each student can expect to lead a discussion at least 2-3 times during the semester.
Students read the paper(s) before coming to class.
Discussion lead reads the papers and works with the instructor to make up a summary of the paper.
During Class
Roughly, each class will proceed as follows:
After Class
Discussion lead writes a short review summarizing the discussion
before sending it out to the class.
Guest Lectures
For some topics, we will have guest lectures from experts who have worked
on the modeling, design and verification closed loop medical devices.
Schedule
The schedule of lectures shown below is subject to change. We will
post papers in the private area for most topics. We will strive to
post all material well in advance. Please take a look through them,
and come prepared for class.
ID | Date | Paper Discussion | Page |
1 | Jan 13th | Introduction - what is a closed-loop medical device? | week 1 |
2 | Jan 20th | Artificial Pancreas: Basic Understanding | week 2 |
3 | Jan 27th | Sriram out of town. AP Survey papers | See week 2 |
4 | Feb 3 | Predictive Pump Shutoff | week 4 |
5 | Feb 10 | PID Controllers for AP | week 5 |
6 | Feb 17 | Mathematical Modeling # 1 | week 6 |
7 | Feb 24 | Mathematical Modeling # 2 | week 7 |
8 | Mar 2 | Model-Predictive Control Basics | week 8 |
9 | Mar 9 | MPC control for artificial pancreas | week 9 |
10 | Mar 16 | Bihormonal control | week 10 |
11 | Mar 30 | Security of insulin pumps | week 11 |
12 | Apr 6 | Human factors for medical devices | week 12 |
13 | Apr 13 | Implantable Defibrillators | week 13 |
14 | Apr 20 | ICD security issues | week 14 |
15 | Apr 27 | Verification of ICDs | week 15
|
Topics
The course will focus on closed loop medical devices.
Closed-loop medical devices involve sophisticated algorithms that
control critical physiological functions through sensing, computation
and actuation. These devices may be completely autonomous, or enjoy a
limited degree of autonomy to assist the human operator.
Examples include:
Artificial Pancreas: Control of insulin infusion to patients with
type-1 diabetes.
Computer Controlled Anesthesia: Control of
anesthesia delivery to surgical patients.
Implantable Heart Defibrillators and Pacemakers: Implanted devices
that automatically control electrical signaling to the human heart,
intervening to prevent dangerous heart conditions.
Other Closed-Loop Devices in Medicine: Feedback control in
surgical robots, deep brain stimulation, closed loop mechanical
ventillation, and emerging closed loop devices.
Closed-loop medical devices are safety-critical: malfunctions often result in serious injury or death to the patient.
The course will primarily focus on research around modeling, design, verification/validation and clinical issues surrounding these devices:
Modeling the relevant aspects of human physiology: insulin-glucose regulatory models, pharamacokinetic/pharmacodynamic models, cardiac modeling (at various scales) and other important mathematical modeling efforts in biomedical engineering.
Design issues surrounding these devices: sensing/actuation limitations, constraints on computation. Control algorithm design will be our main focus.
Verification/Validation
Clinical Evaluation
But these are not the only issues surrounding closed loop devices. In a broader setting, there are many other issues to focus on. The papers we read will touch upon some of these issues as well.
Human Factors Issues (these are often closely tied to control algorithm development).
Security/Privacy Issues.
US FDA regulations, commercialization issues, and so on.
Course Work
Grading is simply through participation. The student is expected to
show up in class, prepared to discuss the paper. When asked to be a
discussion lead, the student should take the time to meet with the
instructor during office hours to ensure that their understanding of
the paper is appropriate for leading the discussion.
|